I hope this makes sense, I'm essentially looking for a matrix in which you can just substitute in the gradient m from y=mx and find the reflected coordinates? If this doesn't make any sense please say why? Regards Tom
2 Answers
The matrix you are searching is:
$$ \dfrac{1}{1+m^2} \left[ \begin{array}{cccc} 1-m^2&2m\\ 2m & m^2-1 \end{array} \right] $$ For a proof see may answer to Linear Transformation with eigenvectors and eigenvalues
- 64,377
-
I believe you have the signs of the lower left and upper right entries wrong. – Rory Daulton Feb 07 '15 at 20:23
-
Yes sir! Tank'you ! I edit end correct my distraction. – Emilio Novati Feb 07 '15 at 20:46
-
Or consider the images of the unit vectors after the transformation. – A-Level Student Dec 19 '20 at 19:00
Your question does make sense, since you describe a linear transformation. Any reflection about a line through the origin is a linear transformation.
One way to find the transformation is to find the image of the point $(1,0)$ [call it $(a,b)$] and the image of the point $(0,1)$ [call it $(c,d)$]. The matrix is then
$$\begin{bmatrix} a & c \\ b & d \\ \end{bmatrix}$$
Here is one way to find $(a,b)$. The line between $(1,0)$ and $(a,b)$ is perpendicular to the line $y=mx$: what does that say about the slope of the line between $(1,0)$ and $(a,b)$? Also, the midpoint of the line segment between $(1,0)$ and $(a,b)$ is on the line $y=mx$.
Each of those two statements gives you a linear equation in $a$ and $b$. Solve those two simultaneous equations and you have your $a$ and $b$. Do almost the same thing to find $c$ and $d$.
There are other ways to find your matrix, of course, but this way uses no advanced concepts.
- 32,888
- 6
- 47
- 64